A recent comparison of neuromelanin and free water (FW) imaging in Parkinson’s disease concluded that FW imaging underperforms. We argue this reflects methodological constraints, not biological truth. Three issues compromise FW sensitivity: use of a registration template that degrades substantia nigra localization, an age imbalance between groups that favors neuromelanin detection, and repetition times insufficient to capture the FW compartment.
With unmet acquisition requirements, the technique was never correctly applied. The recent npj Parkinson’s Disease article comparing neuromelanin MRI and free water (FW) imaging in early Parkinson’s disease (PD)1 raises a fundamental issue regarding the application of appropriate imaging methodology. By concluding that FW imaging does not perform well in early PD, a conclusion solidified in the study’s title, the paper risks framing a methodologically constrained outcome as a biological truth.
The authors acknowledge the use of single-shell clinical diffusion data, but the issue is that the imaging sequence and normalization procedures are limited, as the issue just being single-shell is ill-posed. This distinction is not trivial. FW imaging is not a single entity; it is a family of approaches whose performance is highly dependent on acquisition parameters, modeling strategy, and signal-to-noise considerations.
When these factors are suboptimal, the technique will predictably underperform. In this study, FW values showed no significant differences between early PD and controls, while neuromelanin measures performed well. The absence of a signal should not be conflated with absence of utility.
The methods used in this study illustrate the core issue. There are several issues with the current methods that depart from prior studies of FW in patients with PD: (1) registration template choice, (2) age imbalance between groups, and (3) repetition time (TR) length. The authors used a T1 in MNI space to register the DTI data rather than using a fractional anisotropy (FA) template.
Prior work has demonstrated that registration performance for subcortical structures is significantly inferior when using non-FA templates as the reference, with optimal overlap between hand-drawn and automated regions achieved only when FA templates are used2. This issue is important because in early PD3 and in RBD4 the posterior part of the substantia nigra is where the changes in FW are typically observed. When registration is suboptimal, signal from adjacent subregions bleed together, diluting the focal posterior effect that FW imaging is designed to detect.
Extract — continue reading at the source.
